Apurva Shah1, Jacob Antony Alapatt2, Shweta Prasad3, Jitender Saini4, Pramod Pal3, Ragini Verma2, and Madhura Ingalhalikar1
1Symbiosis Center for Medical Image Analysis, Symbiosis International University, Pune, India, 2Department of Radiology, University of Pennsylvania, Philadelphia, PA, United States, 3Department of Neurology, National Institute of Mental Health and Neurosciences, Bengaluru, India, 4Department of Radiology, National Institute of Mental Health and Neurosciences, Bengaluru, India
Synopsis
Parkinson’s
disease is characterized by degeneration of dopaminergic neurons in
the substantia nigra pars compacta (SNc). To study the
micro-structural and free-water (FW) changes using diffusion-MRI in
the SNc it is critical to extract SNc accurately. Our work employs a
neuromelanin sensitive MRI based atlas to delineate the SNc and
demonstrates significant FW and FW eliminated microstructural
alterations in a large cohort of PD (with and without psychosis) and
its association with PD severity, indicative of novel diagnostic and
progression markers of PD which however demonstrate no role in
genesis of psychosis in PD.
Introduction
Recent
studies have demonstrated elevated levels of freewater (FW) in the
substantia nigra (SN) in Parkinsons’s disease (PD) compared to
healthy controls (HCs)1,2.
The FW estimation is performed using a bi-tensor model that separates
the derived diffusion MRI (dMRI) signal into isotropic (FW) and
tissue-based components3.
In application to PD, the FW is extracted from the small SN region
embedded in the brain-stem, the localization of which is complicated
and is generally performed manually4
or using T2 weighted or susceptibility weighted imaging (SWI).
Manual delineation of SN suffers from low accuracy and
reproducibility while using T2/SWI, as it primarily captures the SN
pars reticulata (SNr), a result of iron sensitivity and not the SN
pars compacta (SNc) which contains the dopaminergic neurons that are
degenerated in PD5.
Our work mitigates this problem by employing a neuromelanin sensitive
MRI based atlas that can accurately delineate the SNc and computes FW
and FW eliminated anisotropy and diffusivity measures using recent
advancements in model initialization6
in a large cohort of PD. Moreover, we also investigate FW variation
with disease severity and based on subtypes of patients with PD only
and PD with psychosis (visual hallucinations).Method
Our
study included 133 subjects with PD and 99 HC that were scanned on a
Philips 3T Achieva MRI scanner. Diffusion-MRI images were acquire
using 32 channel head coil with TR/TE = 8783/62 ms, field of view =
224 × 224 mm, voxel size = 1.75 × 1.75 × 2 mm, and no intersection
gap. DTI was performed along 15 directions with a b value = 1000
s/mm2 and
a single b = 0 s/mm2
image. Pre-processing included skull stripping, motion and eddy
correction (using affine transformation). Free-water computation was
performed using a bi-compartment model with isotropic (FW)
compartment and the tissue compartment. The initialization was based
on log-linear interpolation of mean diffusivity and mean signal from
white matter and CSF from b=0 image facilitating a more stable fit6.
The FW maps and the FW eliminated fractional anisotropy (FAt),
axial diffusivity (AXt)
and radial diffusivity (RDt)
maps were computed. For SN localization, a recent atlas created from
27 controls using NMS-MRI, that accurately delineates the SNc was
employed4.
The regions were mapped to subject space using the Advanced
Normalization Tool (ANTs) deformable registration7.
To examine the group differences in FW levels, FAt,
AXt
and RDt,
multivariate analysis of covariance (mancova) was performed between
HC and PD as well as PD-psychosis and PD without psychosis with age
and gender as covariates and correction for multiple comparisons was
performed using false discovery rates (FDR) with a threshold of
p<0.05. Similarly, test for laterality (R vs. L) was performed on
the four measures in the PD group. To evaluate associations between
the microstructural changes to the severity of the disease, the
residuals from the FW and diffusion measures after regressing out age
and gender were correlated to the Unified Parkinson’s Disease
Rating Scale (UPDRS-III) OFF scores (where available) and duration of
illness (DoI) of patients with PD.Results
Table 1 provides the demographic and
clinical information. Figure
1 displays the SNc region of
interest (probabilistic atlas) that captures the complete nigrosomes
and the nigral matrix. The estimated mean FW in the SNc was
significantly higher in both left (p<0.0001) and right (p<0.0001,
corrected) in PD patients compared to controls with left FW
significantly higher than right in PD patients (p<0.001,
corrected). Interestingly, the FAt
values were also significantly higher in PD patients than in controls
(p<0.005 ,corrected) and the AXt
demonstrated a similar trend (L-SNc: p=0.032, uncorrected, R-SNc:
p=0.001, corrected) while RDt
were higher in controls than in PD (L-SNc: p=0.026, uncorrected)
(Figure 2). Figure 3 illustrates a slice with PD (high FW) and
healthy control (with low FW) in the SNc. The correlation analysis
illustrated a positive correlation between mean FAt
and UPDRS-III OFF scores (p-value=0.019) as well as a trend between
mean FW in right SNc and duration of illness (p-value=0.06) (Figure
4). Finally, no differences in all 4 measures were observed between
the PD without psychosis group and PD-with psychosis.Discussion
Using
accurate delineation of the SNc our work demonstrated that not only
FW but a combination of the diffusion measures (FW, FAt,
AXt
and RDt)
together may provide deeper insights into the PD pathology. Increase
in FW may imply neuro-inflammation in the SNc while increase in FAt
(with increased AXt
and reduced RDt)
may perhaps suggest axonal shrinkage or compression after eliminating
all the FW and concurs with earlier FAt
findings1,2.
Significant asymmetry in FW was indicative of lateralized
degeneration and was in accordance with clinical asymmetry that is
typically observed in PD. Although FW increased with duration of
illness and FAt
with UPDRS, no differences between PD and PD-psychosis suggests that
progression of PD could be independent of onset of psychosis symptoms
although PD-psychosis is typically associated with later stages of
illness. Overall, SNc
diffusion measures could be considered a novel marker for PD and its
progression however it may not play a significant role in the genesis
of psychosis in PD.Acknowledgements
We would like to acknowledge Department of Science and Technology - Science Education and Research Board (DST-SERB) (EMR/2017/004523) for funding to this project. This work was also supported by the National Institutes of Health (NIH) [R01NS096606].References
-
Ofori,
E., Pasternak, O., Planetta, P. J., Burciu, R., Snyder, A., Febo,
M., et al.. (2015). Increased free water in the substantia nigra of
Parkinson's disease: a single-site and multi-site
study. Neurobiology of aging, 36(2), 1097-1104.
- Ofori,
E., Krismer, F., Burciu, R. G., Pasternak, O., McCracken, J. L.,
Lewis, M. M., et al.. (2017). Free water improves detection of
changes in the substantia nigra in parkinsonism: A multisite
study. Movement Disorders, 32(10), 1457-1464.
- Pasternak,
O., Sochen, N., Gur, Y., Intrator, N., & Assaf, Y. (2009). Free
water elimination and mapping from diffusion MRI. Magnetic
Resonance in Medicine: An Official Journal of the International
Society for Magnetic Resonance in Medicine, 62(3),
717-730.
- Safai,
A., Prasad, S., Saini, J., Pal, P. K., &Ingalhalikar, M. (2019).
Microstructural Abnormalities of Substantia Nigra in Parkinson’s
disease: A Neuromelanin Sensitive MRI Atlas Based Study. BioRxiv,
686154.
- Langley,
J., Huddleston, D. E., Chen, X., Sedlacik, J., Zachariah, N., &
Hu, X. (2015). A multicontrast approach for comprehensive imaging of
substantia nigra. Neuroimage, 112, 7-13.
doi:10.1016/j.neuroimage.2015.02.045
- Abdol
Aziz Ould Ismail, Drew Parker, Moises Hernandez-Fernandez, Steven
Brem, Simon Alexander, et al.. Characterizing Peritumoral Tissue
Using Free Water Elimination in Clinical DTI. MICCAI
2018 - 21st International Conference on Medical Image Computing and
Computer Assisted Intervention ; Workshop : Brain Lesion,
Sep 2018, Granada, Spain. pp.1-9.
- Xiao,
Y., Lau, J. C., Anderson, T., DeKraker, J., Collins, D. L., Peters,
T., & Khan, A. R. (2019). Bridging micro and macro: accurate
registration of the BigBrain dataset with the MNI PD25 and ICBM152
atlases. bioRxiv,
561118.